Agentic coding with ClickHouse. One person, one data stack, one full-stack application

A ClickHouse Solutions Architect built a full-stack retail analytics platform called ClickShop in a couple of days using Cursor (AI-powered IDE) and the ClickHouse data stack. The app features real-time dashboards querying billions of rows, 18 specialized AI agents for different business personas (CEO, Sales, Data Analyst), and dual observability layers — infrastructure monitoring via ClickStack and LLM tracing via Langfuse. The architecture combines ClickHouse Cloud for analytics, PostgreSQL for transactions, ClickPipes CDC for real-time sync, and LibreChat for the agentic layer. A key takeaway is that open source components with large public codebases help AI coding tools generate accurate code faster, and that building on a consistent, SQL-based platform reduces architectural complexity significantly.